2.0 Leaderboard
Training data is fixed to 50 demo_clean trajectories × 50 tasks (2,500 demos total) on Aloha-AgileX, then evaluated 100 trials/task under demo_clean and demo_randomized. Use the ranking control to switch titles between Average（c2c+c2r）, clean2random（hard）, and clean2clean（easy）. Co-train and Single-task SFT share one board. The default ranking is Average（c2c+c2r）.
To be listed on this leaderboard, a model must provide publicly released code, publicly released weights, and a technical report (arXiv paper or equivalent public document) describing the method. The code must include post-training and deployment. If the result is fine-tuned from a pretrained model, the pretrained weights must be released as well, so supervised fine-tuning can be rerun from that initialization. To submit results or request listing, contact chentianxing2002@gmail.com with links to the code, weights, and report.
A Unified Standard and Open Ecosystem for Robot Policy Evaluation and Deployment
arXiv:2608.09892
The same adapters serve this RoboTwin 2.0 board and RoboDojo (sim + real). Results here are reproduced via the XPolicyLab standard interface.
| Rank | Method | Depth | Contributor | Average | Clean → Random | Clean → Clean | Track | Date |
|---|
Depth: ✓ requires observed depth or depth-derived point clouds at inference; ✗ does not. Depth estimated from RGB and training-only depth supervision are excluded.
Columns follow the current Average（c2c+c2r） / clean2random（hard） / clean2clean（easy） ranking. The depth filter also applies here. Methods tagged Single are single-task SFT baselines.
XPolicyLab
XPolicyLab bridges RoboTwin and RoboDojo with one policy interface, data format, and train/eval pipeline. Website · Code
Sim Leaderboard
XPolicyLab bridges RoboTwin and RoboDojo: one policy interface, one data format, and one train/eval pipeline. Models listed here can be transferred to RoboDojo (sim + real) without rewriting a line of code. Preview below shows the official Top 10; cells are Score / SR% per capability dimension.
Real Leaderboard
Official RoboDojo real-robot preview (Top 10). Cells are Score / SR% overall and on ARX X5, Piper, and Piper X.
github.com/XPolicyLab/XPolicyLab